Organizations are in a rapid pursuit to implement generative AI, which stands as one of the most pivotal technological advancements in decades. Remarkably, a year and a half after ChatGPT’s launch, which has dramatically transformed our world, 65% of organizations are routinely utilizing AI, as reported by McKinsey. This figure has nearly doubled from the firm’s prior Global Survey conducted just ten months earlier.
Looking forward, the outlook is optimistic: most respondents anticipate that generative AI will bring “significant or disruptive” changes to their industries.
“In 2024, generative AI is no longer seen as a novelty,” stated Alex Singla, senior partner and global co-leader of QuantumBlack, AI by McKinsey. “The potential of the technology is now undisputed. Although most organizations are still in the initial phases of their generative AI journey, we are starting to understand what is effective and what is not in its implementation — and how to create real value with the technology.”
Rapid Increase in AI Investment
Half of the survey respondents indicated that their organizations have integrated AI into at least two business functions, with 67% expecting AI investment to grow over the next three years.
Professional services have seen the most significant rise in AI adoption. Currently, generative AI is primarily utilized in marketing and sales (for content creation, personalization, and generating sales leads); product and service development (for design, scientific literature review, and research); and IT (for help desk chatbots, data management, real-time support, and script suggestions). Additionally, organizations are experiencing the greatest cost savings in human resources.
On average, respondents noted that it takes their organizations one to four months to deploy generative AI into production.
Employees at all levels are becoming more comfortable with AI tools both at work and at home, with many now using generative AI in both professional and personal contexts. Notably, 41% of C-level executives report regularly using generative AI at work.
“The pace of innovation, the emergence of new companies and capabilities, and the surge in investment have been remarkable,” said McKinsey associate partner Bryce Hall. “We’re now observing how leading companies are leveraging these impressive AI and generative AI capabilities to create business value.”
Takers, shapers and makers
McKinsey’s Three Archetypes for Implementing Generative AI
McKinsey identifies three approaches to implementing generative AI: “takers” who utilize off-the-shelf tools; “shapers” who customize these publicly available tools; and “makers” who create their own models from scratch.
The survey revealed that most organizations adopt a mixed approach: approximately 50% of generative AI applications use off-the-shelf tools, while the other half are either significantly customized or developed in-house. This pattern is consistent across various industries, including technology, media, telecommunications, consumer goods, retail, financial services, business, legal, and professional services.
Looking ahead, there will be a shift towards a “buy, build, and partner” strategy, moving away from the traditional “build versus buy” mindset. Organizations will develop ecosystems that integrate proprietary, off-the-shelf, and open-source models, according to Alexander Sukharevsky, senior partner and global co-leader of QuantumBlack, AI by McKinsey.
In the early stages of any technology, adopting straightforward, one-step solutions is common. However, as generative AI becomes more prevalent, this approach is not sustainable. “The future enterprise will depend on a well-coordinated mix of multiple foundational models — combining off-the-shelf solutions with tools finely tuned to specific enterprise needs,” Sukharevsky emphasized.
Challenges with Data, Explainability, and Security
Despite the rapid adoption of generative AI, organizations are not oblivious to its inherent risks. In fact, 44% of respondents have already encountered negative outcomes from their use of generative AI. Common issues include inaccuracies in outputs, cybersecurity vulnerabilities, and challenges in explaining AI decisions. Other concerns involve improper AI utilization, data privacy issues, biases, and potential intellectual property infringements.
McKinsey’s identification of “high performers” underscores specific difficulties related to data. These organizations frequently face issues such as inadequate training data, struggles in defining robust data governance processes, and challenges in swiftly integrating data.
Despite recognizing these challenges, only 18% of respondents reported having an enterprise-wide council or board dedicated to responsible AI governance. Moreover, just one-third identified skills in understanding and mitigating generative AI risks as essential for employees who interact with AI tools. Lareina Yee, senior partner at McKinsey and chair of the McKinsey Technology Council, emphasized the need for responsible AI practices from the outset, stressing the necessity for ongoing education and proactive measures.
Organizations are urged to establish clear governance frameworks and principles for the ethical application of generative AI. This includes implementing safeguards, rigorous training, secure contractual agreements with providers, and educating employees to prevent inadvertent exposure of proprietary data to public models. Incorporating risk management practices into AI development processes is also critical.
“The survey results and our client interactions underscore a growing recognition of responsible AI and a pressing need to implement it effectively,” noted Yee. “Transitioning from awareness to concrete actions will be pivotal moving forward.”